SAM2 Point Segmentation Demo

This project provides a Tkinter-based GUI for interactive point-based segmentation using the SAM2 model.

Features

• Load images and interactively select points for segmentation.
• Run SAM2 segmentation and visualize masks.
• Manage mask history (rename, delete, save).
• Pan (WASD) and zoom image view.
• Select multiple masks to combine into custom masks.

Requirements

• Python
• Pillow
• numpy
• tkinter
• samv2 (pip install git+https://github.com/SauravMaheshkar/samv2.git)

Setup

1. Clone this repository.
2. Install dependencies:

   pip install pillow numpy
   pip install git+https://github.com/SauravMaheshkar/samv2.git

3. Download the SAM2 checkpoints:

   pwsh ./download_checkpoints.ps1

Usage

Run the GUI:

python sam2_select.py

Checkpoints

The script download_checkpoints.ps1 downloads the required model checkpoints:

• sam2_hiera_tiny.pt
• sam2_hiera_small.pt
• sam2_hiera_base_plus.pt
• sam2_hiera_large.pt